TWI766747B - Indicate system and method for cow - Google Patents
Indicate system and method for cow Download PDFInfo
- Publication number
- TWI766747B TWI766747B TW110124225A TW110124225A TWI766747B TW I766747 B TWI766747 B TW I766747B TW 110124225 A TW110124225 A TW 110124225A TW 110124225 A TW110124225 A TW 110124225A TW I766747 B TWI766747 B TW I766747B
- Authority
- TW
- Taiwan
- Prior art keywords
- cow
- image
- cattle
- body temperature
- visible light
- Prior art date
Links
Images
Landscapes
- Image Analysis (AREA)
- Radiation Pyrometers (AREA)
Abstract
Description
本發明係關於一種自動化體溫異常警示系統及方法，尤其是一種非接觸方式量測牛隻體溫的牛隻異常警示系統及方法。 The present invention relates to an automated body temperature abnormality warning system and method, in particular to a cattle abnormality warning system and method for measuring the body temperature of cattle in a non-contact manner.
按，牛隻為恆溫動物，在正常情況下，其體溫是在恆定的範圍內產生細微變化，然而研究指出，牛隻在身體出現不適或疾病時，其體溫會出現異常升高，因此，習知養殖場人員為了能監測並實時記錄牛隻的體溫變化，一般是使用水銀溫度計***牛隻直腸以獲取牛隻溫度，以即時發現、預防和診斷牛隻相關疾病。 Press, cattle are warm-blooded animals. Under normal circumstances, their body temperature changes slightly within a constant range. However, studies have pointed out that when cattle are unwell or sick, their body temperature will rise abnormally. In order to monitor and record the changes in the body temperature of cattle in real time, the staff of the known farms generally use a mercury thermometer to insert the rectum of the cattle to obtain the temperature of the cattle, so as to detect, prevent and diagnose cattle-related diseases in real time.
然而，以水銀溫度計***牛隻直腸的牛隻體溫量測方式，該習知養殖場人員必須先行將牛隻綁定，以避免牛隻產生激烈反應，方可將水銀溫度計***牛隻肛門，以獲取牛隻體溫，目前，是採用養殖場的飲食區域的頭部柵欄將牛隻固定後，再進行直腸體溫量測，該量測方式具有費時費力及執行不易等問題；再且，該量測方式還可能因為水銀溫度計的消毒不確實，而造成牛隻疾病交叉感染的問題。 However, in the method of measuring the body temperature of the cow with a mercury thermometer inserted into the rectum of the cow, the conventional farm personnel must first bind the cow to avoid a violent reaction of the cow, and then insert the mercury thermometer into the anus of the cow to prevent the cow from reacting violently. To obtain the body temperature of cattle, at present, the head fence of the feeding area of the farm is used to fix the cattle, and then the rectal body temperature is measured. This measurement method has the problems of time-consuming, laborious and difficult implementation; The method may also cause the problem of cross-infection of cattle diseases due to the inaccurate disinfection of mercury thermometers.
有鑑於此，有必要提供一種牛隻異常警示系統及方法，以解決上述的問題。 In view of this, it is necessary to provide a cattle abnormality warning system and method to solve the above problems.
為解決上述問題，本發明的目的是提供一種牛隻異常警示系統，係可以透過非接觸方式量測牛隻體溫者。 In order to solve the above problems, the purpose of the present invention is to provide a warning system for cattle abnormality, which can measure the body temperature of cattle through a non-contact method.
本發明的次一目的是提供一種牛隻異常警示系統，係可以將體溫異常的牛隻通報飼養員者。 Another object of the present invention is to provide a cattle abnormal warning system, which can notify the breeder of cattle with abnormal body temperature.
本發明的又一目的是提供一種牛隻異常警示系統，係可以將體溫異常的牛隻與體溫正常的牛隻分群者。 Another object of the present invention is to provide a warning system for cattle abnormality, which can group cattle with abnormal body temperature from cattle with normal body temperature.
本發明的再一目的是提供一種牛隻異常警示方法，係可以達到縮短檢測時間及避免疾病交互感染者。 Another object of the present invention is to provide a method for alerting cattle abnormality, which can shorten the detection time and avoid cross-infection of diseases.
本發明全文所述方向性或其近似用語，例如「前」、「後」、「左」、「右」、「上(頂)」、「下(底)」、「內」、「外」、「側面」等，主要係參考附加圖式的方向，各方向性或其近似用語僅用以輔助說明及理解本發明的各實施例，非用以限制本發明。 The directionality or similar terms used throughout the present disclosure, such as "front", "back", "left", "right", "top (top)", "bottom (bottom)", "inside", "outside" , "side surface", etc., mainly refer to the directions of the attached drawings, each directionality or its similar terms are only used to assist the description and understanding of the various embodiments of the present invention, and are not intended to limit the present invention.
本發明全文所記載的元件及構件使用「一」或「一個」之量詞，僅是為了方便使用且提供本發明範圍的通常意義；於本發明中應被解讀為包括一個或至少一個，且單一的概念也包括複數的情況，除非其明顯意指其他意思。 The use of the quantifier "a" or "an" for the elements and components described throughout the present invention is only for convenience and provides a general meaning of the scope of the present invention; in the present invention, it should be construed as including one or at least one, and a single The concept of also includes the plural case unless it is obvious that it means otherwise.
本發明全文所述之「資料庫單元(Database Unit)」，係指將一群相關的電子資料集合並儲存在硬碟、記憶體或上述之組合，且可藉由資料庫管理系統(DBSMS)所提供的語法功能，例如新增、讀取、搜尋、更新及刪除等，對電子資料進行相關處理；該資料庫管理系統可以藉由不同資料結構方式管理電子資料，例如可以為關聯式、階層式、網狀式或物件導向式等，本發明係以關聯式資料庫管理系統為例進行以下說明，惟非用以限制本發明。 The "Database Unit" mentioned in the whole text of the present invention refers to a group of related electronic data that is collected and stored in a hard disk, a memory or a combination of the above, and can be managed by a database management system (DBSMS). Provides grammatical functions, such as adding, reading, searching, updating and deleting, etc., to process electronic data; the database management system can manage electronic data through different data structures, such as relational, hierarchical , mesh type or object-oriented type, etc., the present invention is described below by taking a relational database management system as an example, but is not intended to limit the present invention.
本發明全文所述之「耦接(Coupling)」，係指二裝置之間可 藉由任何直接或間接的連接手段，以相互傳遞資料。舉例而言，第一裝置耦接第二裝置，於本發明中應被解讀為該第一裝置可以直接連接該第二裝置，例如可以藉由有線實體(如：電線、排線、走線、雙絞線)連接；或者該第一裝置可以透過其他裝置或某種連接手段而間接地連接該第二裝置，例如可以藉由無線媒介(如：WiFi、藍芽)或異質網路(Heterogeneous Network)連接，本領域中具有通常知識者可以依據欲相連之裝置的常態連接手段予以選擇者。 "Coupling" as used throughout the present invention refers to the connection between the two devices. By any direct or indirect means of connection, to transfer data to each other. For example, the first device is coupled to the second device, which should be construed in the present invention as the first device can be directly connected to the second device, for example, through wired entities (such as wires, cables, wiring, Twisted pair) connection; or the first device can be indirectly connected to the second device through other devices or some connection means, such as wireless media (such as WiFi, Bluetooth) or a heterogeneous network (Heterogeneous Network) ) connection, a person with ordinary knowledge in the art can select it according to the normal connection means of the device to be connected.
本發明的牛隻異常警示系統，包含：一資料庫單元，用以儲存數頭牛隻各自的識別碼及正常體溫；一物體偵測單元，用以產生一感應區域，並偵測是否有一牛隻進入該感應區域，以產生一阻斷訊號；一成像模組，用以朝進入該感應區域的該牛隻拍攝，以產生一可見光影像及一熱影像，該可見光影像及該熱影像分別包含該牛隻身上的一待檢測部位，該成像模組將該可見光影像與該熱影像執行影像融合，以獲得一融合影像，且由該融合影像中獲得該牛隻的待檢測部位的當下體溫；及一檢驗平台，耦接該資料庫單元、該物體偵測單元及該成像模組，該檢驗平台接收到該阻斷訊號後，控制該成像模組作動，以獲得該可見光影像，以及該牛隻的待檢測部位的當下體溫，該檢驗平台將該可見光影像輸入至一牛臉辨識模組，以辨識出該牛隻的一識別碼，並根據該識別碼由該資料庫單元中取得該牛隻的待檢測部位的正常體溫，該檢驗平台比對該當下體溫是否大於該正常體溫，且該當下體溫與該正常體溫的溫差是否超出一溫差門檻，若比對結果為是，該檢驗平台發出一體溫異常訊號至一行動裝置，使持有該行動裝置的飼養員得知該牛隻的體溫出現異常。 The cattle abnormality warning system of the present invention includes: a database unit for storing the respective identification codes and normal body temperature of several cattle; an object detection unit for generating a sensing area and detecting whether there is a cattle Only enter the sensing area to generate a blocking signal; an imaging module is used to photograph the cattle entering the sensing area to generate a visible light image and a thermal image, the visible light image and the thermal image respectively include For a part to be detected on the cow, the imaging module performs image fusion of the visible light image and the thermal image to obtain a fused image, and obtains the current body temperature of the part to be detected of the cow from the fused image; and an inspection platform, which is coupled to the database unit, the object detection unit and the imaging module. After the inspection platform receives the blocking signal, it controls the imaging module to act to obtain the visible light image and the imaging module. the current body temperature of the part to be detected, the inspection platform inputs the visible light image to a cow face recognition module to identify an identification code of the cow, and obtains the cow from the database unit according to the identification code Only the normal body temperature of the part to be detected, the test platform compares whether the current body temperature is greater than the normal body temperature, and whether the temperature difference between the current body temperature and the normal body temperature exceeds a temperature difference threshold, if the comparison result is yes, the test platform issues An abnormal body temperature signal is sent to a mobile device, so that the breeder who holds the mobile device knows that the body temperature of the cow is abnormal.
本發明的牛隻異常警示方法，包含：偵測是否有一牛隻進入一感應區域，以產生一阻斷訊號；在接收到該阻斷訊號後，朝進入該感應區域 的該牛隻拍攝，以產生一可見光影像及一熱影像，該可見光影像及該熱影像分別包含該牛隻身上的一待檢測部位；將該可見光影像與該熱影像執行影像融合，以獲得一融合影像；由該融合影像中獲得該牛隻的待檢測部位的當下體溫；透過深度學習技術辨識該可見光影像，以辨識出該牛隻的一識別碼；根據該識別碼由一資料庫中取得該牛隻的待檢測部位的正常體溫；及比對該當下體溫是否大於該正常體溫，且該當下體溫與該正常體溫的溫差是否超出一溫差門檻，若比對結果為是，則發送包含該識別碼的一體溫異常訊號至一行動裝置，使持有該行動裝置的飼養員得知該牛隻的體溫出現異常。 The cow abnormal warning method of the present invention includes: detecting whether a cow enters an induction area to generate a blocking signal; after receiving the blocking signal, entering the induction area The cow is photographed to generate a visible light image and a thermal image, the visible light image and the thermal image respectively include a part to be detected on the cow; image fusion is performed on the visible light image and the thermal image to obtain a fused image; obtain the current body temperature of the part to be detected of the cow from the fused image; identify the visible light image through deep learning technology to identify an identification code of the cow; obtain from a database according to the identification code The normal body temperature of the to-be-detected part of the cow; and compare whether the current body temperature is greater than the normal body temperature, and whether the temperature difference between the current body temperature and the normal body temperature exceeds a temperature difference threshold, if the comparison result is yes, send a message containing the An abnormal body temperature signal of the identification code is sent to a mobile device, so that the breeder who holds the mobile device knows that the body temperature of the cow is abnormal.
據此，本發明的牛隻異常警示系統及方法，係可以透過熱成像儀，以非接觸方式朝牛隻的待檢測部位拍攝，以獲得該牛隻的待檢測部位的當下體溫，並將該當下體溫與該牛隻在未感染疾病時的正常體溫互相比對，若該當下體溫大於該正常體溫且溫差超出一溫差門檻時，則發送一體溫異常訊號至飼養員的行動裝置，使該飼養員可以得知該牛隻的體溫出現異常。如此，本發明的牛隻異常警示系統及方法，係可以達到縮短檢測時間、降低檢測難度及避免疾病交互感染的功效。 Accordingly, the cattle abnormality warning system and method of the present invention can use a thermal imager to photograph the part to be detected of the cow in a non-contact manner, so as to obtain the current body temperature of the part to be detected of the cow, and record the current temperature of the part to be detected. The lower body temperature is compared with the normal body temperature of the cow when it is not infected with the disease. If the current body temperature is greater than the normal body temperature and the temperature difference exceeds a temperature difference threshold, an abnormal body temperature signal will be sent to the breeder's mobile device, so that the breeding The staff can know that the body temperature of the cow is abnormal. In this way, the cattle abnormality warning system and method of the present invention can achieve the effects of shortening the detection time, reducing the detection difficulty and avoiding the cross infection of diseases.
其中，該成像模組由該可見光影像中取得一感興趣區域，該感興趣區域內具有該待檢測部位的影像，該成像模組根據一變換矩陣，以由該熱影像中取得相對應該感興趣區域相對位置的影像資料，再將該相對位置的影像資料與該可見光影像的該感興趣區域進行重疊，以獲得該融合影像。如此，係具有簡單運算以避免增加系統運算負擔的功效。 Wherein, the imaging module obtains a region of interest from the visible light image, and the region of interest has an image of the part to be detected, and the imaging module obtains a region of interest from the thermal image according to a transformation matrix The image data of the relative position of the region is then overlapped with the region of interest of the visible light image to obtain the fusion image. In this way, the system has the effect of simple operation to avoid increasing the system operation burden.
其中，該牛臉辨識模組係可以採用卷積神經網路，該牛臉辨識模型的訓練資料係可以包含數頭牛隻各自的臉部影像，該臉部影像係可以包含該牛隻的眼睛、鼻子及耳朵等身體部位，以及釘掛於該牛隻耳朵上的耳標。如此，係具有自動辨識出牛隻身分的功效。 Wherein, the cow face recognition module can use a convolutional neural network, the training data of the cow face recognition model can include the respective facial images of several cows, and the facial images can include the eyes of the cows , body parts such as nose and ears, as well as ear tags that are attached to the cow's ears. In this way, the system has the effect of automatically identifying the cow's identity.
其中，該訓練資料係可以具有該數頭牛隻各自身上的花紋影像。如此，係具有提升辨識出牛隻身分準確率的功效。 Wherein, the training data may have images of patterns on each of the several cows. In this way, the system has the effect of improving the accuracy of identifying the cow's identity.
本發明的牛隻異常警示系統，還可以另包含一分群設備，該分群設備具有一分群通道，該分群通道具有一入口端及一出口端，該分群通道用以限制該牛隻由該入口端往該出口端的方向移動，該出口端可以具有一第一柵門及一第二柵門，該第一柵門與該第二柵門分別耦接該檢驗平台，當該檢驗平台的比對結果為是時，係可以控制該第一柵門開啟，使該牛隻通過該第一柵門來到一體溫異常區域，反之，係可以控制該第二柵門開啟，使該牛隻通過該第二柵門來到一體溫正常區域，該體溫異常區域與該體溫正常區域互不連通。如此，係可以將體溫異常的牛隻與體溫正常的牛隻分群，係具有避免牛隻疾病交互感染的功效。 The cattle abnormality warning system of the present invention may further include a grouping device, the grouping device has a grouping channel, the grouping channel has an inlet end and an outlet end, and the grouping channel is used to restrict the cattle from passing from the inlet end. Move towards the direction of the exit end, the exit end may have a first gate and a second gate, the first gate and the second gate are respectively coupled to the inspection platform, when the comparison result of the inspection platform When it is true, the system can control the opening of the first gate so that the cow passes through the first gate to an area with abnormal body temperature; otherwise, the system can control the opening of the second gate so that the cow passes through the first gate. The second gate came to a normal body temperature area, and the abnormal body temperature area and the normal body temperature area were not connected to each other. In this way, cattle with abnormal body temperature can be grouped with cattle with normal body temperature, which has the effect of avoiding cross infection of cattle diseases.
其中，由該可見光影像中取得一感興趣區域，該感興趣區域內具有該待檢測部位的影像，根據一變換矩陣由該熱影像中取得相對應該感興趣區域相對位置的影像資料，再將該相對位置的影像資料與該可見光影像的該感興趣區域進行重疊，以獲得該融合影像。如此，係具有簡單運算以避免增加系統運算負擔的功效。 Wherein, a region of interest is obtained from the visible light image, and the region of interest has the image of the part to be detected, and the image data corresponding to the relative position of the region of interest is obtained from the thermal image according to a transformation matrix, and then the image data is obtained from the thermal image. The image data at the relative position is overlapped with the region of interest of the visible light image to obtain the fusion image. In this way, the system has the effect of simple operation to avoid increasing the system operation burden.
其中，係可以透過深度學習技術建立一牛臉辨識模型，該牛臉辨識模型的訓練資料係可以包含數頭牛隻各自的臉部影像，該臉部影像包含該牛隻的眼睛、鼻子及耳朵等身體部位，以及釘掛於該牛隻耳朵上的耳標。如此，係具有自動辨識出牛隻身分的功效。 Among them, a cow face recognition model can be established through deep learning technology, and the training data of the cow face recognition model can include the respective facial images of several cows, and the facial images include the eyes, nose and ears of the cow and other body parts, as well as ear tags that are attached to the cow's ears. In this way, the system has the effect of automatically identifying the cow's identity.
其中，該訓練資料係可以具有該數頭牛隻各自身上的花紋影像。如此，係具有提升辨識出牛隻身分準確率的功效。 Wherein, the training data may have images of patterns on each of the several cows. In this way, the system has the effect of improving the accuracy of identifying the cow's identity.
其中，係可以使該牛隻位於一分群通道中，該分群通道的出口端設置二閘門，根據比對結果係可以發送相對應控制指令到一控制器，並透 過該控制器驅動相對應的閘門開啟，以將該牛隻送入到相對應的區域，該二區域互不連通。如此，係可以將體溫異常的牛隻與體溫正常的牛隻分群，係具有避免牛隻疾病交互感染的功效。 Among them, the system can make the cattle located in a grouping channel, and the exit end of the grouping channel is provided with two gates. According to the comparison result, the system can send corresponding control commands to a controller, and transparently The controller drives the corresponding gate to open, so as to send the cattle into the corresponding area, and the two areas are not connected to each other. In this way, cattle with abnormal body temperature can be grouped with cattle with normal body temperature, which has the effect of avoiding cross infection of cattle diseases.
〔本發明〕 〔this invention〕
1:資料庫單元 1: Database unit
2:物體偵測單元 2: Object detection unit
3:成像模組 3: Imaging module
31:可見光攝影機 31: Visible light camera
32:熱成像儀 32: Thermal Imager
4:檢驗平台 4: Inspection platform
41:牛臉辨識模組 41: Cow face recognition module
5:分群設備 5: Grouping equipment
51:分群通道 51: Grouping channel
52:第一柵門 52: The first gate
53:第二柵門 53: Second gate
A1:體溫異常區域 A1: Area with abnormal body temperature
A2:體溫正常區域 A2: Area with normal body temperature
C:牛隻 C: cattle
E1:入口端 E1: entry port
E2:出口端 E2: Exit port
S1:偵測步驟 S1: Detection step
S2:成像步驟 S2: Imaging step
S21:拍攝步驟 S21: Shooting steps
S22:融合步驟 S22: Fusion step
S3:數據比對步驟 S3: Data comparison step
S4:分群步驟 S4: Grouping step
〔第1圖〕本發明一較佳實施例的系統方塊圖。 [FIG. 1] A system block diagram of a preferred embodiment of the present invention.
〔第2圖〕本發明一較佳實施例的系統架構圖。 [FIG. 2] A system architecture diagram of a preferred embodiment of the present invention.
〔第3圖〕本發明一較佳實施例的方法流程圖。 [FIG. 3] A flow chart of a method according to a preferred embodiment of the present invention.
為讓本發明之上述及其他目的、特徵及優點能更明顯易懂，下文特舉本發明之較佳實施例，並配合所附圖式，作詳細說明如下： In order to make the above-mentioned and other objects, features and advantages of the present invention more obvious and easy to understand, the preferred embodiments of the present invention are exemplified below, and are described in detail as follows in conjunction with the accompanying drawings:
請參照第1圖所示，其係本發明牛隻異常警示系統的一較佳實施例，係包含一資料庫單元1、一物體偵測單元2、一成像模組3及一檢驗平台4，該資料庫單元1、該物體偵測單元2及該成像模組3耦接該檢驗平台4。
Please refer to FIG. 1, which is a preferred embodiment of the cattle abnormality warning system of the present invention, which includes a
該資料庫單元1用以儲存數頭牛隻各自的識別碼及正常體溫，該識別碼用以作為各該牛隻的身分證明，在本實施例中，該識別碼可以為該牛隻的耳標號碼(ear tag)或隨機產生的流水號(serial number)；該正常體溫為各該牛隻在未感染疾病時，其身上的一待檢測部位的歷史平均體溫，在本實施例中，該待檢測部位可以分別為牛隻的眼窩周圍、***或蹄部，並用以作為評估是否警示獸醫師牛隻可能罹患流行熱、***炎及蹄病等疾病的要件之一。
The
該物體偵測單元2用以產生一感應區域，並偵測是否有一牛隻進入該感應區域，以產生一阻斷訊號，在本實施例中，該物體偵測單元2可以為一阻斷式紅外線感應器，該阻斷式紅外線感應器具有一發射器及一接收器。該發射器與該接收器對位設置，且由該發射器朝該接收器發送一紅外線。當該牛隻位於該發射器與該接收器之間時，該紅外線被該牛隻阻斷，使該接收器無法接收到該紅外線，進而使該物體偵測單元2產生該阻斷訊號。
The
該成像模組3用以朝進入該感應區域的該牛隻拍攝，以產生一可見光影像(visible image)及一熱影像(thermal image)，該可見光影像及該熱影像分別包含該牛隻身上的一待檢測部位，在本實施例中，係可以透過一可見光攝影機31及一熱成像儀32分別朝該牛隻拍攝，以產生該可見光影像及該熱影像，例如但不限制地，該可見光影像與該熱影像的影像尺寸可以皆為640X480。值得一提的是，當該待檢測部位為該牛隻的***或蹄部時，該可見光攝影機31的數量可以為兩台，以分別拍攝該牛隻的臉部及***/蹄部，為本領域具有通常知識者可以理解。該成像模組3將該可見光影像與該熱影像執行影像融合(Image Fusion)，以獲得一融合影像，且由該融合影像中獲得該牛隻的待檢測部位的當下體溫，在本實施例中，該當下體溫可以為該融合影像的影像陣列中的當下最高溫度。
The
具體而言，該成像模組3係可以由該可見光影像中取得一感興趣區域(ROI)，該感興趣區域內具有該待檢測部位的影像；該成像模組3係可以根據一變換矩陣(Perspective Transformation)，以由該熱影像中取得相對應該感興趣區域相對位置的影像資料；該成像模組3再將該相對位置的影像資料與該可見光影像的該感興趣區域進行重疊，以獲得該融合影像。該變換矩陣的公式係可以如下式(1)~(3)所示：
Specifically, the
該檢驗平台4耦接該資料庫單元1、該物體偵測單元2及該成像模組3，在本實施例中，該檢驗平台4係可以採用一樹莓派(Raspberry Pi 3/4)作為控制平台。該檢驗平台4接收到該阻斷訊號後，控制該成像模組3作動，以獲得該可見光影像，以及該牛隻的待檢測部位的當下體溫；該檢驗平台4將該可見光影像輸入至一牛臉辨識模組41，以辨識出該牛隻的一識別碼，並根據該識別碼由該資料庫單元1中取得該牛隻的待檢測部位的正常體溫；該檢驗平台4比對該當下體溫是否大於該正常體溫，且該當下體溫與該正常體溫的溫差是否超出一溫差門檻(例如：該溫差門檻可以為2℃)，若比對結果為是，該檢驗平台4發出一體溫異常訊號至一行動裝置，使持有該行動裝置的飼養員得知該牛隻的體溫出現異常；若比對結果為否，則可以不需執行額外作動。
The
承上述，係可以透過卷積神經網路(CNN)建立該牛臉辨識模型41，並將該可見光影像輸入至該牛臉辨識模型41中，以辨識出該識別碼。具體而言，係可以透過卷積神經網路建立一基礎模型，該基礎模型具有一輸
入層(Input Layer)、一隱藏層(Hidden Layer)及一輸出層(Output Layer)。其中，該隱藏層係可以由連續6組卷積層(Convolution Layer)與池化層(Pooling Layer)，以及另外7個卷積層所組成，該基礎模型係可以採用YOLOv3-tiny網路架構完成，在本實施例中，該輸入層及該輸出層的神經元(Neuron)數量，可以分別為1及15個，該輸出層的輸出結果係為數頭牛隻各自的識別碼的機率值，並以具有最高機率值者作為最終的輸出結果。舉例而言，耳標號碼為1的機率值為0.913，耳標號碼為2的機率值為0.693，則最終的輸出結果即為該牛隻的識別碼為1。
Based on the above, the cow
其中，該基礎模型的訓練方式可以為監督式學習，且其訓練影像的來源可以包含數頭牛隻各自的臉部影像，該臉部影像包含該牛隻的眼睛、鼻子及耳朵等身體部位，以及釘掛於該牛隻耳朵上的耳標，較佳地，該訓練影像的來源還可以具有該數頭牛隻各自身上的花紋影像；再且，在該基礎模型的初次訓練時，該數頭牛隻的數量可以為15頭，各該牛隻的臉部影像及花紋影像的影像數量至少需要3152張，以輔助辨識出各該牛隻的識別碼。藉此，該基礎模型可以自動學習如何辨識出牛隻的耳標號碼。上述影像張數僅為範例，而非作為本發明的限制。值得一提的是，當牛隻耳朵上未釘掛耳標時，係可以設定該輸出層的輸出結果的機率值若均小於一機率門檻時，則可以自動產生一流水號作為該牛隻的識別碼，係本發明所屬技術領域中具有通常知識者可以理解。 Wherein, the training method of the basic model may be supervised learning, and the source of the training images may include respective face images of several cows, and the face images include the body parts such as the eyes, nose and ears of the cow, and the ear tag nailed on the cow's ear, preferably, the source of the training image can also have pattern images on each of the several cows; moreover, during the initial training of the basic model, the data The number of cows can be 15, and the number of images of each cow's face and pattern images needs to be at least 3152 to assist in identifying the identification code of each cow. In this way, the basic model can automatically learn how to recognize the ear tag number of a cow. The above-mentioned number of images is only an example, and not a limitation of the present invention. It is worth mentioning that when there is no ear tag on the cow's ear, the system can set the probability value of the output result of the output layer. If the probability value of the output result is less than a probability threshold, the first water number can be automatically generated as the cow's The identification code can be understood by those with ordinary knowledge in the technical field to which the present invention pertains.
其中，該訓練影像可以切割成訓練資料及測試資料，更進一步地，該訓練資料還可以切割出驗證資料，以對該基礎模型進行訓練與驗證，以調整該基礎模型的神經網路的權重。 The training image can be cut into training data and test data, and further, the training data can also be cut into verification data, so as to train and verify the basic model and adjust the weight of the neural network of the basic model.
該基礎模型以該訓練資料進行訓練時，係可以由該訓練資料中隨機取得一批量大小(BatchSize)的資料，以決定該基礎模型中的每個神經
元的權重值，以建立一候選神經元路徑。隨後，該輸入層由該訓練資料中，隨機取得另一批量大小的資料，以建立另一候選神經元路徑，直到該訓練資料中的所有資料都已輸入至該基礎模型中，或是該候選神經元路徑的數量達到一預設門檻即可停止。再者，重覆執行上述程序500,200次(Iteration)，並評估該數個候選神經元路徑各自的預測能力，並以具有最好預測能力的候選神經元路徑作為該牛臉辨識模型41實際使用的深度神經元路徑，據此，該牛臉辨識模型41的準確度可以來到89.5%。
When the basic model is trained with the training data, a batch size (BatchSize) can be randomly obtained from the training data to determine each neuron in the basic model
The weight value of the element to establish a candidate neuron path. Then, the input layer randomly obtains another batch size of data from the training data to create another candidate neuron path until all the data in the training data has been input into the base model, or the candidate neuron path It stops when the number of neuron paths reaches a preset threshold. Furthermore, the above procedure is repeated 500,200 times (Iteration), and the respective prediction capabilities of the several candidate neuron paths are evaluated, and the candidate neuron path with the best prediction ability is used as the actual use of the cow
本發明牛隻異常警示系統，還可以具有一分群設備5，該分群設備5具有一分群通道51，該分群通道51具有一入口端E1及一出口端E2，該分群通道51用以限制該牛隻僅能由該入口端E1往該出口端E2的方向移動，在本實施例中，該分群通道51的寬度無法同時提供二頭牛並排行走；又，該物體偵測單元2及該成像模組3係分別位於該分群通道51旁，且相較於該出口端E2，係較鄰近該入口端E1設置。該出口端E2具有一第一柵門52及一第二柵門53，該第一柵門52與該第二柵門53分別耦接該檢驗平台4，當該檢驗平台4的比對結果為是時，控制該第一柵門52開啟，使該牛隻通過該第一柵門52來到一體溫異常區域A1，並將該牛隻歸類為體溫異常對象；反之，控制該第二柵門53開啟，使該牛隻通過該第二柵門53來到一體溫正常區域A2，並將該牛隻歸類為體溫正常對象，該體溫異常區域A1與該體溫正常區域A2互不連通，舉例而言，係可以透過圍籬各自圈繞出該體溫異常區域A1與該體溫正常區域A2，惟不以此為限。
The cattle abnormality warning system of the present invention can also have a
請參照第2圖所示，本發明的牛隻疾病系統應用於檢測牛隻的流行熱疾病時，牛隻的待檢測部位係為其眼窩周圍；位於牧場的數頭牛隻C分別被引導進入該分群通道51內，當進入至該分群通道51內的牛隻C遮斷由該物體偵測單元2所發出的紅外線時，該物體偵測單元2產生一阻斷訊號
至該檢驗平台4，使該檢驗平台4驅動該成像模組3的可見光攝影機31及熱成像儀32，分別對該牛隻C的眼窩周圍進行拍攝，以產生一可見光影像及一熱影像。該成像模組3將該可見光影像與該熱影像執行影像融合，以獲得一融合影像，且由該融合影像中獲得該牛隻C的眼窩周圍的當下體溫。
Referring to Figure 2, when the cattle disease system of the present invention is applied to detect epidemic diseases in cattle, the part to be detected of the cattle is around the eye socket; several cattle C located in the pasture are guided into the In the
該檢驗平台4將該可見光影像輸入至該牛臉辨識模組41，以辨識出該牛隻C的一識別碼，並根據該識別碼由該資料庫單元1中取得該牛隻C在未感染流行熱疾病時，其眼窩周圍的正常體溫；該檢驗平台4比對該當下體溫是否大於該正常體溫，且該當下體溫與該正常體溫的溫差超出2℃，若比對結果為是，該檢驗平台4控制該第一柵門52開啟，使該牛隻C通過該第一柵門52來到一體溫異常區域A1，並將該牛隻C歸類為體溫異常對象，且發出一體溫異常訊號至一行動裝置，使持有該行動裝置的飼養員得知該牛隻C的體溫出現異常；反之，該檢驗平台4控制該第二柵門53開啟，使該牛隻C通過該第二柵門53來到一體溫正常區域A2，並將該牛隻C歸類為體溫正常對象。
The
請參照第3圖所示，其係本發明牛隻異常警示方法的一較佳實施例，係包含一偵測步驟S1、一成像步驟S2及一數據比對步驟S3。 Please refer to FIG. 3 , which is a preferred embodiment of the cattle abnormality warning method of the present invention, which includes a detection step S1 , an imaging step S2 and a data comparison step S3 .
該偵測步驟S1係偵測是否有牛隻進入一感應區域，以產生一阻斷訊號，在本實施例中，係可以透過上述物體偵測單元2，以產生該感應區域。
The detecting step S1 is to detect whether a cow enters a sensing area, so as to generate a blocking signal. In this embodiment, the sensing area can be generated through the above-mentioned
該成像步驟S2可以分為一拍攝步驟S21及一融合步驟S22，該拍攝步驟S21係在接收到該阻斷訊號後，朝進入該感應區域的該牛隻拍攝，以產生一可見光影像及一熱影像，該可見光影像及該熱影像分別包含該牛隻身上的一待檢測部位，在本實施例中，係可以透過上述可見光攝影機31及上述熱成像儀32分別朝該牛隻拍攝，以產生該可見光影像及該熱影像；此外，
該待檢測部位可以為該牛隻的眼窩周圍、***及蹄部，並用以作為評估是否警示獸醫師牛隻可能罹患流行熱、***炎及蹄病等疾病的要件之一。
The imaging step S2 can be divided into a photographing step S21 and a fusion step S22. The photographing step S21 is to photograph the cow entering the sensing area after receiving the blocking signal to generate a visible light image and a thermal image image, the visible light image and the thermal image respectively include a to-be-detected part of the cow. In this embodiment, the
該融合步驟S22係將該可見光影像與該熱影像執行影像融合，以獲得一融合影像，且由該融合影像中獲得該牛隻的待檢測部位的當下體溫。具體而言，該融合步驟S22係可以由該可見光影像中取得一感興趣區域，該感興趣區域內具有該待檢測部位的影像；根據一變換矩陣由該熱影像中取得相對應該感興趣區域相對位置的影像資料，再將該相對位置的影像資料與該可見光影像的該感興趣區域進行重疊，以獲得該融合影像，該變換矩陣係可以如上述公式(1)~(3)所示。 The fusion step S22 is to perform image fusion of the visible light image and the thermal image to obtain a fusion image, and obtain the current body temperature of the to-be-detected part of the cow from the fusion image. Specifically, in the fusion step S22, a region of interest can be obtained from the visible light image, and the region of interest has an image of the to-be-detected portion; and the relative region of interest is obtained from the thermal image according to a transformation matrix. position image data, and then overlap the relative position image data with the region of interest of the visible light image to obtain the fusion image, and the transformation matrix can be as shown in the above formulas (1) to (3).
該數據比對步驟S3係透過深度學習技術(Deep Learning)產生上述牛臉辨識模型41辨識該可見光影像，以辨識出該牛隻的一識別碼，該識別碼用以作為該牛隻的身分證明，在本實施例中，該識別碼可以為該牛隻的耳標號碼或隨機產生的流水號。再者，根據該識別碼由一資料庫中取得該牛隻的待檢測部位的正常體溫，在本實施例中，該正常體溫可以為該牛隻在未感染疾病時，其身上的待檢測部位的歷史平均體溫。隨後，比對該當下體溫是否大於該正常體溫，且該當下體溫與該正常體溫的溫差是否超出一溫差門檻(例如：該溫差門檻可以為2℃)，若比對結果為是，則發送包含該識別碼的一體溫異常訊號至一行動裝置，使持有該行動裝置的飼養員得知該牛隻的體溫出現異常；若比對結果為否，則可以不需執行額外作動。
The data comparison step S3 is to generate the above-mentioned cow
本發明牛隻異常警示方法，還可以具有一分群步驟S4，該分群步驟S4係使該牛隻位於一分群通道中，該分群通道的出口端係可以設置二閘門；根據比對結果發送相對應控制指令到一控制器，並透過該控制器驅動相對應的閘門開啟，以將該牛隻送入到相對應的區域，該二區域互不連通，在本實施例中，其中一閘門所對應的區域係供體溫正常的牛隻停留，另一閘 門所對應的區域係供體溫異常的牛隻停留。值得一提的是，例如可以透過一圍籬圈繞出該二區域，惟不以此為限。 The cattle abnormality warning method of the present invention may also have a grouping step S4, the grouping step S4 is to make the cattle be located in a grouping channel, and two gates may be set at the outlet end of the grouping channel; according to the comparison result, corresponding The control command is sent to a controller, and the corresponding gate is driven to open through the controller, so as to send the cattle to the corresponding area. The two areas are not connected to each other. In this embodiment, one of the gates corresponds to The area is reserved for normal body temperature cattle, and the other gate The area corresponding to the door is reserved for cattle with abnormal body temperature. It is worth mentioning that, for example, the two areas can be surrounded by a fence, but not limited thereto.
綜上所述，本發明的牛隻異常警示系統及方法，係可以透過熱成像儀，以非接觸方式朝牛隻的待檢測部位拍攝，以獲得該牛隻的待檢測部位的當下體溫，並將該當下體溫與該牛隻在未感染疾病時的正常體溫互相比對，若該當下體溫大於該正常體溫且溫差超出一溫差門檻時，則發送一體溫異常訊號至飼養員的行動裝置，使持有該飼養員可以得知該牛隻的體溫出現異常。如此，本發明的牛隻異常警示系統及方法，係可以達到縮短檢測時間、降低檢測難度及避免疾病交互感染的功效。 To sum up, the cattle abnormality warning system and method of the present invention can use a thermal imager to photograph the part to be detected of the cow in a non-contact manner, so as to obtain the current body temperature of the part to be detected of the cow, and The current body temperature is compared with the normal body temperature of the cow when it is not infected with the disease. If the current body temperature is greater than the normal body temperature and the temperature difference exceeds a temperature difference threshold, an abnormal body temperature signal is sent to the breeder's mobile device, so that Holding the breeder can know that the body temperature of the cow is abnormal. In this way, the cattle abnormality warning system and method of the present invention can achieve the effects of shortening the detection time, reducing the detection difficulty and avoiding the cross infection of diseases.
雖然本發明已利用上述較佳實施例揭示，然其並非用以限定本發明，任何熟習此技藝者在不脫離本發明之精神和範圍之內，相對上述實施例進行各種更動與修改仍屬本發明所保護之技術範疇，因此本發明之保護範圍當視後附之申請專利範圍所界定者為準。 Although the present invention has been disclosed by the above-mentioned preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make various changes and modifications relative to the above-mentioned embodiments without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the patent application attached hereto.
1:資料庫單元 1: Database unit
2:物體偵測單元 2: Object detection unit
3:成像模組 3: Imaging module
31:可見光攝影機 31: Visible light camera
32:熱成像儀 32: Thermal Imager
4:檢驗平台 4: Inspection platform
41:牛臉辨識模組 41: Cow face recognition module
5:分群設備 5: Grouping equipment
52:第一柵門 52: The first gate
53:第二柵門 53: Second gate
Claims (10)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| TW110124225A TWI766747B (en) | 2021-07-01 | 2021-07-01 | Indicate system and method for cow |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| TW110124225A TWI766747B (en) | 2021-07-01 | 2021-07-01 | Indicate system and method for cow |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| TWI766747B true TWI766747B (en) | 2022-06-01 |
| TW202303536A TW202303536A (en) | 2023-01-16 |
Family
ID=83103791
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| TW110124225A TWI766747B (en) | 2021-07-01 | 2021-07-01 | Indicate system and method for cow |
Country Status (1)
| Country | Link |
|---|---|
| TW (1) | TWI766747B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN209432306U (en) * | 2019-02-01 | 2019-09-24 | 湖北俏牛儿牧业有限公司 | A kind of certainly mobile ox Thermometer System |
| TWM604939U (en) * | 2020-06-30 | 2020-12-01 | 樹德科技大學 | Cow medical care device |
| CN112468766A (en) * | 2020-03-31 | 2021-03-09 | 丰疆智能软件科技(南京)有限公司 | Cattle state monitoring system and method |
-
2021
- 2021-07-01 TW TW110124225A patent/TWI766747B/en active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN209432306U (en) * | 2019-02-01 | 2019-09-24 | 湖北俏牛儿牧业有限公司 | A kind of certainly mobile ox Thermometer System |
| CN112468766A (en) * | 2020-03-31 | 2021-03-09 | 丰疆智能软件科技(南京)有限公司 | Cattle state monitoring system and method |
| TWM604939U (en) * | 2020-06-30 | 2020-12-01 | 樹德科技大學 | Cow medical care device |
Non-Patent Citations (2)
| Title |
|---|
| 陳怡璇、陳宜鴻、蕭方君,"智慧養牛科技：乳牛健康一把罩─紅外線體溫感測",<https://scitechvista.nat.gov.tw/Article/C000003/detail?ID=2f9603b1-a983-476e-9b42-dc0a2ae400af>,科技大觀園,公開日期：106年2月3日 * |
| 陳怡璇、陳宜鴻、蕭方君,"智慧養牛科技：乳牛健康一把罩─紅外線體溫感測",<https://scitechvista.nat.gov.tw/Article/C000003/detail?ID=2f9603b1-a983-476e-9b42-dc0a2ae400af>,科技大觀園,公開日期：106年2月3日。 |
Also Published As
| Publication number | Publication date |
|---|---|
| TW202303536A (en) | 2023-01-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| AU2013350867B2 (en) | Systems and methods for predicting the outcome of a state of a subject | |
| WO2021227350A1 (en) | Method and apparatus for measuring temperature, electronic device, and computer-readable storage medium | |
| JP6203238B2 (en) | Livestock management system | |
| JP7460798B2 (en) | Smart livestock management system and method | |
| US9740921B2 (en) | Image processing sensor systems | |
| US20110043630A1 (en) | Image Processing Sensor Systems | |
| Hitelman et al. | Biometric identification of sheep via a machine-vision system | |
| JPWO2018025404A1 (en) | Diagnostic equipment | |
| KR102527058B1 (en) | Apparatus for detecting mounting behavior of cattle | |
| KR20140119228A (en) | Smart Terminal Liaison type Real Time Medical Examination Method and Total Management System thereof | |
| CN111898580A (en) | System, method and equipment for acquiring body temperature and respiration data of people wearing masks | |
| CN111666826A (en) | Method, apparatus, electronic device and computer-readable storage medium for processing image | |
| CN109784200A (en) | Milk cow behavior image based on binocular vision obtains and body condition intelligent monitor system | |
| CN107019498A (en) | Nurse robot | |
| KR101445698B1 (en) | Real Time Auto Medical Examination Method and Domestic Animal Auto Care System thereof | |
| Noe et al. | Automatic detection and tracking of mounting behavior in cattle using a deep learning-based instance segmentation model | |
| TWI766747B (en) | Indicate system and method for cow | |
| CA3175199A1 (en) | Apparatus and methodologies for improved detection of important biological states in animals | |
| WO2018143889A1 (en) | Systems and methods for determining likelihood of states in cattle animal | |
| Jaddoa et al. | Automatic temperature measurement for hot spots in face region of cattle using infrared thermography | |
| KR20130088667A (en) | U-it based remote image medical device for improvement of livestock productivity | |
| KR102372107B1 (en) | Image-based sow farrowing notification system | |
| CN107485412A (en) | A kind of ox health monitoring systems and method | |
| Yuan et al. | Stress-free detection technologies for pig growth based on welfare farming: A review | |
| KR102516271B1 (en) | Animal management system using iris recognition on long distance |